P-79

Mumbai High Field Heterogeneity Addressed Through Quantitative

Seismic Reservoir Characterization
S. Dasgupta*, S. Volterrani*, R. Bachrach*, A. Carrillat*, S. Sharma*, N. Dutta*, M. Pratap**,
R Bhanja**, A. Saha**, and T. Friedel*, *Schlumberger DCS ** ONGC Ltd.
Summary
We present a case study of heterogeneous carbonate reservoir characterization conducted via quantitative seismic reservoir
analysis of pre-stack inversion with rock physics modelling and Bayesian joint stochastic inversion. The jointly predicted
porosity and volume of clay are then used as input for static reservoir modeling.

Introduction derive reservoir and overburden properties from surface

The Mumbai High field is a brown field located 160km
WNW off Mumbai city in India. The field consists in a
doubly plunging asymmetric anticlinal structure with a
gentle western limb. The sedimentary sequence is of
Tertiary age and most productive reservoirs of the field are
within the Miocene carbonate LII and LIII units. The main
challenges in Mumbai High field are reservoir
heterogeneity and excessive gas production from the large
gas cap. Reservoir heterogeneity has complicated the
understanding of the water front within the reservoirs
making it difficult to locate bypassed oil and optimally
distribute water injection for efficient flooding. Reservoir
heterogeneity arises from the variation in
porosity/permeability in the carbonate layers affected by
changes in clay content and marked diagenetic overprint
and dissolution. This study focuses on reservoir
heterogeneity mapping from quantitative seismic reservoir
characterization. The analysis is based on pre-stack
inversion and rock physics modeling.
Objectives
This seismic reservoir characterization analysis is based on
seismic inversion data and rock physics, which is part of
larger study covering static and dynamic full field
modeling.
The main objective of the seismic inversion and rock
physics analysis is to support reservoir characterization and
quantitative interpretation using continuous seismic data to
seismic. The quantitative seismic reservoir characterization
aims at providing:
• Continuous data for static model building and flow
simulation.
• Support of infill well target selection process.
• Risk analysis for new wells
In this approach, rock physics is the key enabling
technology that links seismic properties to reservoir
properties.
Methodology
This analysis is based on combining pre-stack seismic
inversion and rock physics modeling. As such, the work
scope was subdivided into 5 steps:
1. The initial data preparatory work for facilitating the
pre-stack inversion. It includes data editing for all available
vertical wells with measured elastic properties
(compressional and shear velocity, density) and shear
velocity prediction for the vertical well with only
compressional velocity measurement.
2. Well data analysis for establishing the rock physics
model. Rock physics model describes the relationship
between the elastic properties such as bulk modulus and
shear modulus, density and reservoir properties like
porosity, saturation and volume of clay.
3. Seismic pre-stack inversion. The inversion workflow
was designed to maximize the resolution by broadening the
bandwidth and correcting the seismic wavelet to zero

Data and Consulting Services , Schlumberger Asia Services Limited, NSE Estate, Western Express Highway, Goregaon(E),
Mumbai-400067, India, SDasgupta@bombay.oilfield.slb.com
 Mumbai High Field Heterogeneity

phase. It also aimed at obtaining high resolution seismic

acoustic impedance from broad band zero phase data. The
primary objective of this phase was to generate an acoustic
impedance volume from the 3D depth migrated seismic
volume (PSDM) of sufficiently good quality such that
reservoir properties derived from it (i.e., Phit) could be
successfully used to populate the reservoir static model.
4. Stochastic rock simulation from the rock physics
model and calculation of elastic properties form the
stochastically simulated properties (porosity, Vclay,
saturation).
5. Generation of litho-cube and joint stochastic inversion
for prediction of reservoir properties with associated
probabilities.

The method is based on a complete Bayesian approach that

integrates different measurements at different scale. It uses
Bayesian classification techniques (Sengupta and Bachrach,
2007) for generating the probability density functions
(PDFs) associated with different lithology. The advantage
of this approach, compared to other classical inversion
methods, is to ensure consistency between properties, here
porosity and volume of clay prediction, by running a joint
estimation of both properties together using a rock physics
model. The stochastic simulation and Bayesian estimation
theory enable to address the multi-scales effect between log
domain and seismic domain and to capture associated
uncertainties. The volume of clay/porosity inversion is
based on rock-physics relationships and well-log
calibration. The stochastic simulation addresses the Figure 1: Modified Hashin–Shtrickman model jointly predict
uncertainty of the elastic/stiffness relationship for a given boundary of the data for bulk (top) and shear (bottom) moduli
reservoir property pair (porosity/Vclay pair) and vice-versa versus porosity.
(Bachrach, 2006).
Results
The exploratory rock physics analysis shows that clay
content within the reservoir has strong effect on moduli-
porosity trends for both bulk and shear data (Figure 1). The
rock model reproduces main aspect of reservoir properties;
porosity trends with respect to clay content. Rock physics
analysis also suggests that clay content is more significant
than saturation effects. The seismic pre-stack inversion was
run on the PSDM 3D data over the 1750 sq. km covering
the field area. Though the general quality of the seismic
gathers is not optimum for AVO inversion, good quality P
Impedance and PR (Poisson’s ratio) were obtained via
extensive data preconditioning of the input gathers.

2
 Mumbai High Field Heterogeneity

Subsequent reservoir characterization is performed by
combining the inverted P Impedance and PR with the rock
physics model through Bayesian joint stochastic inversion,
which results in good porosity and volume of clay
prediction throughout the field.
Bayesian joint stochastic inversion results in very good
porosity and volume of clay and prediction correlate with
well measured property. Correlation coefficients are above
71% (R2) for Vclay and above 62% (R2) for porosity for
both L-III and L-II reservoirs.

As part of the analysis of the results, a full validation
exercise is conducted separately both qualitatively and
quantitatively. From a qualitative standpoint, visual
comparison of petrophysically derived and joint inversion
predicted porosity and volume of clay for all wells used in
NW
0
Scale (m)
400 800
Figure 2: Jointly inverted Vclay Maximum A Posteriori Probability
overlaid with Vclay log at well location
the inversion shows good tie (Figure 2). Correlation
coefficient analysis (R2) is conducted on blind wells
between porosity and volume of clay curves from the wells
and the seismic elastic derived porosity and volume of clay
outputs. The correlation analysis is taken as a measure of
the quality of the joint stochastic inversion results.
Finally, the joint inversion results are used for static
reservoir property modelling of the field that is used to
guide ongoing infill well drilling campaigns.
References
Bachrach, R., 2006. Joint estimation of porosity and
SE
saturation using stochastic rock-physics modeling.
Geophysics, 71, O56-O53.
Sengupta, M. and Bachrach, R., 2007. Uncertainty in
seismic-based pay volume estimation: Analysis using rock
physics and Bayesian statistics. The Leading Edge. 184-
189.
Statement from Author
View expressed in the paper are those of authors only and
can be different from the official version of Oil and Natural
Gas Corporation Ltd. Data utilized and referred for the
paper will not affect ONGC’s Business interest in any way.
Acknowledgments
The authors are thankful to Oil and Natural Gas
Corporation Ltd for permitting the publication of the paper.

Conclusions
Rock physics analysis shows that clay content within
reservoir has strong effect on moduli-porosity trends both
for bulk and shear data in Mumbai High field. Rock model
reproduces main aspect of reservoir properties; porosity
trends with respect to clay content. However, saturation
effects are found within range of model uncertainty i.e.
masked by the model uncertainty, and clay content is more
significant than saturation effects. Contribution of small
variation in Vclay is higher than saturation changes.